Opportunities

Research Scientist Positions

We are looking for research scientists to conduct novel research in AI for healthcare at IHU Strasbourg/University of Strasbourg in the areas of medical image analysis, computer aided surgery, surgical data science and artificial intelligence. More information is available here.

Research Engineers Positions

We are have positions open for research engineers with background in medical image analysis, computer aided surgery, surgical data science or federated learning interested to develop and support our prototypes. More information is available here. Please also check out these two new federated learning positions.

Fellowships for Clinicians

We have positions for clinicians interested to join our team and work closely with our researchers on AI projects during 6-24 months. Please send an email to Nicolas Padoy with a letter of motivation, curriculum vitae and list of publications if interested. More information about the general IHU fellowship program is available here.

Team Lead for a Medical AI Service Platform

We have one position open for an experienced team lead to develop a Medical Artificial Intelligence service platform at IHU Strasbourg. Proficiency in French is expected. More information here.

PhD and Post-doctoral Positions

In the context of the national AI Chair project AI4ORSafety, which aims at developing new AI methods for automating the assessment of critical safety steps during surgery, we are currently looking for two PhD students and one post-doctoral fellow in Computer Vision/Artificial Intelligence:

(1) Phd position: Federated Learning for Scaling-up Surgical Activity Analysis
(2) Phd position: Self-supervision for Anatomy and Activity Recognition in Endoscopic Videos
(3) Post-doctoral position: AI Methods for the Automated Assessment of Critical Safety Steps in Surgery

Internships

We are offering several internship positions / Master’s theses for motivated students interested in contributing to the development of our computer vision system for the operating room. Proficiency in programming and English along with knowledge in computer vision or machine learning are expected. If you are interested in doing an internship in an area different from the ones listed below, please send Nicolas Padoy a short email including a description of your interests, curriculum vitae and academic transcripts. All positions are funded and have a duration from 5 to 6 months. Successful candidates will be hosted within the IHU institute at the University Hospital of Strasbourg. He/She will thereby have direct contact with clinicians and also have access to an exceptional research environment.

To apply: please send an email to the listed contact describing your interests and motivation along with your curriculum vitae and academic transcripts.

Topics:

Deep Learning for Activity Recognition in Large Surgical Video Databases
	Develop an approach to model, detect and recognize the surgical activities occurring within the hundreds of videos from our constantly growing surgical database. The approach will focus either on endoscopic videos or on multiple external RGBD views capturing the clinicians' activities. It will also be integrated into our live demonstrator for the operating room. (contact)
Multi-view Human Body Tracking / Activity Analysis for the Operating Room
	Design a robust method to estimate the 3D body poses of clinicians or recognize their activities using a multi-RGBD camera system mounted directly in the operating room of our clinical partner. The approach will have to address several issues related to clutter, occlusions and temporal tracking. (contact)
Anatomy Detection Algorithms from Endoscopic Ultrasound
	Propose new algorithms for the analysis of Endoscopic Ultrasound (EUS) videos. The project requires developing computer vision and machine learning algorithms for the analysis of EUS videos of pancreatic anatomy. Viable solutions are likely to transfer to clinical practice and actively assist clinicians on Ultrasound based anatomy recognition. (contact)
3D Visualization of X-ray Propagation for Radiation Safety Analysis
	Propose a method for the intuitive visualization of the 3D radiation risk in the interventional radiology room using the Microsoft HoloLens head-mounted display. The method will rely on our ceiling-mounted multi-RGBD camera system and be implemented on GPU to enable quick interactions between the clinicians and the monitoring system. (contact)
Computer Vision for Histopathological Diagnosis
	Pathology is the gold standard for the understanding of micro-biological phenomena based on the examination of body tissues. The project considers developing algorithms for the analysis of samples of cells to identify structural properties of the tissue. To that end, we use image segmentation algorithms to identify cells and graph-based algorithms for the analysis of their geometric structure. (contact)
Explainable approaches to Computed Tomography based diagnosis
	Computed Tomography (CT) scans offer very rich information source for clinical diagnosis. We develop data-efficient explainable algorithms for diagnostic analysis of CT scans. We consider explicit models of data symmetries to obtain low-dimensional invariant representations that maintain sufficient information to be discriminative while simplifying the learning task.
Raspberry Pi based Multi-view Camera System for Real-time Deep Learning Activity Analysis in Operating Room
	Development of an electronic board system based on Raspberry Pi allowing the acquisition of synchronised video data streams from RGB-D (Intel RealSense) cameras. This system will make it possible to obtain the 3D spatial positions of the various elements of an operating room scene such as the clinical staff and the surgical systems (respirator, endoscopy columns, ...) using state-of-the-art neural network detection algorithms. (contact)
Machine Learning for Medical Image Analysis
	Today’s operating room (OR) has been transformed into a convoluted setting of machines, surgeons, nurses, and patients. Large amounts of data are generated just in a single operation. These data are temporal and multimodal (e.g. endoscopy videos, radiological, physiological, human movement, etc.) providing a rich context of the operation. In this project, the intern will research self-supervision , weak supervision, multimodal fusion methods to segment, classify or analyze medical images such as CT and MRI scans. (contact)
Deep Learning based Ultra-Fast X Ray Simulation Tool in Operating Room
	Study of an innovative approach to simulate the X-ray flow during hybrid surgery. During this type of procedure, the surgeon is guided by an X-ray system (fluoroscopy or radioscopy), which exposes him/her to ionising radiation. Real-time simulation of this radiation is the only way to know the real exposure of the medical staff, but classical simulation methods (Monte Carlo) are not fast enough. We propose an alternative approach combining PCA and neural network to obtain a real-time radiation map to minimize the impact on practitioners. (contact)
Development of Intelligent Annotation Functions in a Web-based Collaborative Tool and Application to Deep Learning Video Analysis
	In order to produce high quality medical databases, the annotation of elements such as videos or images is an essential but unfortunately extremely time consuming task. We are developing a web-based collaborative solution (MOSaiC) that allows surgeons and endoscopists to efficiently participate in this crucial phase for artificial intelligence systems. The development of functions to improve this efficiency (semi-automated segmentation, intelligent annotation tools, etc.) will reduce annotation time while ensuring the necessary quality. (contact)
Federated Learning Methods for Medical Images and Videos
	Federated Learning (FL) is a new technique that has been proposed to circumvent concerns related to privacy and data ownership during machine learning. Our lab is developing infrastructure for rich temporal multi-modal data in the operating room, and we are developing methods that can help us improve efficiency of FL algorithms. The intern will have the opportunity to work with a group of scientists and clinicians and research novel algorithms related to privacy preserving approaches, noisy data and self-supervision in FL settings. (contact)

More details and other topics available upon request. Illustrative videos related to several projects are available here.